Pancreatic ductal adenocarcinoma (PDAC) is the fourth leading cause of cancer death in the United States. The exceptionally poor prognosis of PDAC can be largely attributed to the difficulty of detecting the cancer at an early stage when curative surgical resection remains a viable option. The less than 10% patients who present with small, surgically resectable cancers have a realistic chance of cure and a 5-year survival rate of 20-30%. The identification of serum biomarkers associated with a high probability of PDAC could substantially change clinical practice by changing the risk-reward ratio for invasive imaging procedures or facilitating the development of less invasive procedures. At present, the overriding goal for the early detection of PDAC is the reliable identification of resectable tumors of less than 1.0 cm in size with negative lymph nodes. Presently, there are no biomarkers able to capture the disease more than 1 year prior to symptoms onset where there is a high probability of detecting PDAC at early stages. Biomarkers developed using samples obtained from patients at diagnosis do not validate well in preclinical samples indicating that different se of biomarkers characterizes early asymptomatic PDAC than late and symptomatic disease where biomarkers of inflammation and acute phase could predominate. We thus hypothesize that to generate a robust algorithm for identification of resectable tumors of less than 1.0 cm in size with negative lymph nodes, biomarker discovery should be performed in preclinical samples. We further hypothesize that biomarker velocity should be considered in order to generate optimized classification algorithm. We have generated a 5-biomarker classification algorithm that diagnoses patients 12-24 months before diagnosis with 743% SN at 98% SP and 24-35 months before diagnosis - with 64% SN at 98% SP. We have additionally accumulated promising preliminary data obtained in individual Women Health Initiative and pooled PLCO samples identifying biomarkers that are differentially expressed in preclinical cases vs. healthy controls and that demonstrate time-to-diagnoses dependent changes, thus providing support for our hypothesis that pre-clinical samples will be useful for discovery of pre-clinical biomarkers We propose to validate these candidate biomarkers in individual PLCO samples including longitudinal samples, develop an optimized classification algorithm, and validate this algorithm in preclinical samples from two other screening clinical trials. We propose following Specific Aims: 1. Optimize and validate the performance of preclinical PDAC classification algorithm in prospective sample. 2. Validate performance of a pre- diagnostic classification panel in two large prospective cohorts. 3. Construct Risk of Pancreatic Malignancy (RPM) algorithm to add the velocity component to PDAC classification. At the completion of the proposed study we expect to have identified a set of biomarkers with individual robust performance in preclinical PDAC samples, and have generated a bioinformatics algorithm based on biomarker velocities for recognition of PDAC at the resectable stages with high sensitivity and specificity.